Evapotranspiration (ET) is the loss of water from soil and plants to the atmosphere through evaporation and transpiration. Accurate ET estimation is crucial for water resource management, particularly in agriculture. Traditional methods, like the FAO-56 Penman-Monteith method, require complex meteorological data that may not always be available. Recently, remote sensing (RS) and machine learning (ML) techniques, such as Random Forest Regression (RFR), have become effective alternatives for estimating ET, even in data-scarce regions. This work evaluates the RFR model for predicting ET in Sinaloa, Mexico, using only maximum and minimum temperature data. The model was trained on records from 1980 to 2010 and evaluated on data from 2011 to 2018. Training samples were obtained from the TerraClimate dataset, and the results were compared with Hargreaves estimates. The model achieved a coefficient of determination (r2) of 0.932 and a Root Mean Square Error (RMSE) of 0.471, surpassing the Hargreaves method by 12.56% and 15.44%, respectively. This ML + RS approach provides a promising alternative to traditional methods, improving the reliability of water management and irrigation planning in data-scarce scenarios.